SELECT语句中的DB2 SQL合并行

Question

I have the following query for which I would like to merge rows for BusinessPhone, Fax, mobile number, email. At the moment these results are returned on separate lines. How can I achieve this?

SELECT DISTINCT
ABAN8 as AddressNumber,
(CASE WHEN TRIM(ALADD2)!='' THEN CONCAT(CONCAT(TRIM(ALADD1),', '),(CASE WHEN     TRIM(ALADD3)!='' THEN CONCAT(CONCAT(TRIM(ALADD2),', '),(CASE WHEN TRIM(ALADD4)!='' THEN CONCAT(CONCAT(TRIM(ALADD3),', '), TRIM(ALADD4)) ELSE TRIM(ALADD3) END)) ELSE TRIM(ALADD2) END)) ELSE TRIM(ALADD1) END) AS Address,
TRIM(ALCTY1) as City,
TRIM(ALCOUN) as Country,
TRIM(ALADDZ) as PostCode,
TRIM(ABALPH) as Company,
TRIM(WWATTL) as JobTitle,
TRIM(WWGNNM) as FirstName,
TRIM(WWSRNM) as Surname,
CASE WHEN WPPHTP = 'COM' THEN WPPH1 END BusinessPhone,
CASE WHEN WPPHTP = 'FACS' THEN WPPH1 END AS FAX,
CASE WHEN WPPHTP = 'MOB' THEN WPPH1 END AS MobileNumber,
CASE WHEN WPPHTP = 'HOME' THEN WPPH1 END AS HomePhone,
CASE WHEN WPPHTP = 'EML' THEN WPPH1 END AS EmailAddress
FROM CLTDTA.F0101 
LEFT OUTER JOIN CLTDTA.F0111 ON ABAN8 = WWAN8 
LEFT OUTER JOIN CLTDTA.F0115 ON ABAN8 = WPAN8
LEFT OUTER JOIN CLTDTA.F0116 ON ABAN8 = ALAN8
WHERE 
(CLTDTA.F0101.ABAT1 = 'ST' OR CLTDTA.F0101.ABAT1 = 'SC')

Answer 1

To solve this one, you get to play games with the joins. To start with, I'm going to make the following assumptions based on your sample query:

• Your primary key to the F0101 table is the address number (ABAN8). It is unique.
• Each address type ('ST' or 'SC') has its own unique address number (ABAN8).
• You only want one row per address number value.
• Phone numbers and email addresses are stored in F0115. There is a one-to-many relationship between F0101 and F0115.
• Address lines are stored in F0115. There is a one-to-one relationship between F0101 and F0115.

With the one-to-many relationship between F0101 and F0115, a simple join query results in multiple address records being returned: one for each phone/email record. But if we change the join so that it only returns one row, and do a separate join for each phone type, then we can treat the phone numbers (and email) as separate fields in a single row. So, let's remove the LEFT OUTER JOIN CLTDTA.F0115 ON ABAN8=WPAN8 and replace it with five joins:

LEFT OUTER JOIN CLTDTA.F0115 ON ABAN8 = WPAN8 AND WPPHTP = 'COM' AS COM115
LEFT OUTER JOIN CLTDTA.F0115 ON ABAN8 = WPAN8 AND WPPHTP = 'FACS' AS FAX115
LEFT OUTER JOIN CLTDTA.F0115 ON ABAN8 = WPAN8 AND WPPHTP = 'MOB' AS MOB115
LEFT OUTER JOIN CLTDTA.F0115 ON ABAN8 = WPAN8 AND WPPHTP = 'HOME' AS HOME115
LEFT OUTER JOIN CLTDTA.F0115 ON ABAN8 = WPAN8 AND WPPHTP = 'EML' AS EML115

Be aware that some of these joins will fail, returning null values, if there isn't a certain phone type for a certain address. This is normal, expected, and why we use left joins instead of inner joins.

To finish things out, we change out the part of the SELECT clause with all of those CASEs and just replace them with the fields in question, pulling from the appropriate join. This is why the table aliases were used in the joins, you have to be able to specify which of the 5 versions of F0115 we want to pull a particular value from:

COM115.WPPH1 AS BusinessPhone,
FACS115.WPPH1 AS FAX,
MOB115.WPPH1 AS MobileNumber,
HOME115.WPPH1 AS HomePhone,
EML115.WPPH1 AS EmailAddress

Depending on your application, if returning nulls is a problem, you can wrap each of those WPPH1 fields in an IFNULL() function. Here is the final query:

SELECT DISTINCT
    ABAN8 as AddressNumber,
    (CASE WHEN TRIM(ALADD2)!='' THEN CONCAT(CONCAT(TRIM(ALADD1),', '),(CASE WHEN     TRIM(ALADD3)!='' THEN CONCAT(CONCAT(TRIM(ALADD2),', '),(CASE WHEN TRIM(ALADD4)!='' THEN CONCAT(CONCAT(TRIM(ALADD3),', '), TRIM(ALADD4)) ELSE TRIM(ALADD3) END)) ELSE TRIM(ALADD2) END)) ELSE TRIM(ALADD1) END) AS Address,
    TRIM(ALCTY1) as City,
    TRIM(ALCOUN) as Country,
    TRIM(ALADDZ) as PostCode,
    TRIM(ABALPH) as Company,
    TRIM(WWATTL) as JobTitle,
    TRIM(WWGNNM) as FirstName,
    TRIM(WWSRNM) as Surname,
    IFNULL(COM115.WPPH1) AS BusinessPhone,
    IFNULL(FACS115.WPPH1) AS FAX,
    IFNULL(MOB115.WPPH1) AS MobileNumber,
    IFNULL(HOME115.WPPH1) AS HomePhone,
    IFNULL(EML115.WPPH1) AS EmailAddress
FROM CLTDTA.F0101 
LEFT OUTER JOIN CLTDTA.F0111 ON ABAN8 = WWAN8 
LEFT OUTER JOIN CLTDTA.F0115 ON ABAN8 = WPAN8 AND WPPHTP = 'COM' AS COM115
LEFT OUTER JOIN CLTDTA.F0115 ON ABAN8 = WPAN8 AND WPPHTP = 'FACS' AS FAX115
LEFT OUTER JOIN CLTDTA.F0115 ON ABAN8 = WPAN8 AND WPPHTP = 'MOB' AS MOB115
LEFT OUTER JOIN CLTDTA.F0115 ON ABAN8 = WPAN8 AND WPPHTP = 'HOME' AS HOME115
LEFT OUTER JOIN CLTDTA.F0115 ON ABAN8 = WPAN8 AND WPPHTP = 'EML' AS EML115
LEFT OUTER JOIN CLTDTA.F0116 ON ABAN8 = ALAN8
WHERE (CLTDTA.F0101.ABAT1 = 'ST' OR CLTDTA.F0101.ABAT1 = 'SC')

Answer 2

See jarlh answer in the comments for solution.

You have to GROUP BY all selected columns that are not arguments to set functions. (And do MAX on the CASE expressions.